# Image volumes are stable in 1.36. Anyone mounting model weights with them yet? — retained record

> Retained reddit record from reddit:kubernetes.

Canonical: https://highsignal.app/data/records/73a0b954de76313c

[← sources](https://highsignal.app/data)


source: [reddit](https://highsignal.app/data/reddit) (reddit:kubernetes)  published 2026-11-03 15:25 ingested 2026-11-04 02:40

[original source ↗](https://www.reddit.com/r/kubernetes/comments/1xmvvae/image_volumes_are_stable_in_136_anyone_mounting/)

## retained text



```
With image volumes now GA, you can mount an OCI artifact read-only into a pod, so the serving image and the model version separately. On paper that kills the init container that pulls weights from S3. I help maintain KitOps, which packages models as OCI artifacts (ModelKits) with weights, config, and adapters as separate layers. 

Most people run it as an init container that does `kit unpack`. Image volumes could make that step disappear, so I'm trying to learn what actually works before we lean on it. For anyone who's tried it: which runtimes handled non-image OCI artifacts for you, and which forced you to wrap the model as a fake image? How did 20 replicas cold-starting at once behave? Are you pinning by digest, or does a tag move trigger a rollout? Mostly want to hear what broke.
```



document fetched 2026-11-04 00:36 · canonical https://reddit.com/r/kubernetes/comments/1xmvvae/image_volumes_are_stable_in_136_anyone_mounting
